Meta’s TRIBE v2: The AI Model That Predicts Brain Activity and Why It Could Change Everything

Meta has developed TRIBE v2, a multimodal AI model designed to predict how the human brain responds to various sensory inputs like text, audio, and video. Unlike traditional neuroscience models that focus on a single specialized task, this system uses transformer architecture to simulate integrated cognitive processing across broad regions of the brain. This tool acts as a scientific instrument, allowing researchers to pilot experiments virtually and test hypotheses before conducting expensive fMRI scans. While it is an encoding model that predicts reactions rather than a decoding tool that reads private thoughts, it represents a significant shift toward foundation models in neuroscience. By aligning AI more closely with human perception, this technology holds potential for advancements in healthcare, brain-computer interfaces, and the design of more intuitive digital products. Ultimately, the research signals a transition where the brain is treated as a predictive computational layer, similar to how AI has transformed the study of proteins and language.